Complications and pitfalls of lumbar interlaminar and transforaminal epidural injections

Infections have been reported to occur in 1–2% of spinal injections [22]. Severe infections are noted to be rare with an incidence of 0.1–0.01% of all spinal injections [22]. They vary between meningitis [23, 24], epidural abscess [23, 25, 26], osteomyelitis [27, 28], and discitis [29]. Staphylococcus aureus is the most common organism reported to be found. It is believed to be introduced via the skin through needle puncture. It is usually introduced due to poor sterile technique. Undetected and untreated infection can lead to sepsis and spread to other sites through direct contiguous spread or through Batson’s plexus [22]. Neurological deficits can occur due to compression from exudate [30]. Patients with immunocompromising conditions such as diabetes are more susceptible to infection and should be followed closely [30].

Infection from gram-negative anaerobes can theoretically occur by unintentional penetration into the intestinal or pelvic cavity [22]. This is particularly easy to do in S1 TFESIs. The needle can unintentionally go through the dorsal foramen and past the ventral foramen entering the pelvic cavity (Fig. 3). In Fig. 3a, the needle is placed using “tunnel vision” into the S1 dorsal foramen. With this technique there may be a lack of bony structure to stop needle advancement. Lateral imaging (Fig. 3b) is critical to check needle depth as it is relatively easy to advance past the ventral foramen and into the pelvic cavity. If this should occur, i.e., the needle depth is miscalculated prior to going to lateral imaging, we recommend removing the needle completely and discarding it from the sterile field rather than repositioning the needle more posteriorly.

In the other lumbar TFESIs there is potential to go past the foramen, along the lateral border of the vertebral body, and into the abdominal cavity potentially piercing the intestinal cavity. At times in the subpedicular or retroneural approaches the SAP obstructs the path to the “6 o’clock” position on the pedicle. The needle is advanced laterally to avoid the SAP. If advanced too far lateral the needle tip can end up along the lateral aspect of the vertebral body instead of more medial into the intervertebral foramen. Lateral imaging helps to further confirm the depth of the needle. If the needle depth goes past the anterior border of the vertebral body, it can potentially pierce the intestinal cavity. In the primary author’s experience, this rarely occurs. However; one must be alerted to this possibility. Once again, if the needle happened to be too far anterior, we recommend discarding the needle rather than repositioning more posterior and medially to minimize the chance of introducing iatrogenic infection.

Piercing a vessel is an inherent risk to all injection procedures. Potential complications associated with this are bleeding, the formation of hematomas, and intravascular injection of medication. The potential for bleeding and hematoma formation is increased in patients with a coagulopathy, liver disease, or in patients that take anticoagulant medications such as warfarin or clopidogrel [22, 31]. Damage to the underlying vessels may lead to hematomas that cannot be visualized under traditional fluoroscopy.

Patients use chronic anticoagulant therapy for the prevention of myocardial ischemia or stroke, thromboprophylaxis after surgery, or treatment of acute thromboembolism [32]. The intensity and duration of anticoagulation affect the risk of spontaneous, as well as procedural-related spinal bleeding [32, 33]. This risk is increased significantly with the use of multiple anticoagulants [32, 33]. The following guidelines for performing spinal procedures in anticoagulated patients are based on the second American Society of Regional Anesthesia and Pain Medicine (ASRA) Consensus Conference on Neuraxial Anesthesia and Anticoagulation in 2003 [32, 34]. Warfarin therapy should be discontinued 4–5 days before spinal procedures and the international normalized ratio (INR) should be within normal range at the time of the procedure to ensure adequate levels of all vitamin K-dependent factors [32, 34]. Thienopyridine derivatives, e.g., clopidogrel, ticlopidine, should be suspended 7 days and 14 days, respectively, prior to spinal procedures to allow for recovery of primary and secondary platelet aggregation and platelet–fibrinogen binding [32, 34]. Aspirin and non-steroidal antiinflammatory drugs (NSAIDs) have not been found to have any contraindications for spinal procedures [32, 34, 35]. Low-molecular weight heparin (LMWH) should be held for at least 12 h before the procedure in thromboprophylactic dosing and at least 24 h in therapeutic dosing [32, 34].

The incidence of epidural hematoma is approximated to be less than 1 in 150,000 epidurals [34]. The actual incidence of neurologic dysfunction resulting from hemorrhagic complications is unknown [34]. Epidural hematomas can lead to compression of the spinal nerve or nerves causing irreversible damage [36]. Nerve injury may be minimized by evacuation of the hematoma within 24 h of the first symptoms [37]. In the lead author’s experience, an epidural hematoma was incidentally discovered while performing an ILESI (Fig. 4). The patient had a myelogram 2 weeks prior to the epidural injection at a different facility, which presumably led to the unrecognized epidural hematoma. During the ILESI the needle was unintentionally placed in the superior aspect of the hematoma. Upon injection of contrast, an unusual dye spread occurred. Because the pattern of spread could not be definitively distinguished, the ILESI was aborted and a TFESI was performed instead. Later, a magnetic resonance imaging (MRI) of the lumbar spine was obtained demonstrating the epidural hematoma. The contrast spread, in essence, is an epidural hematogram. Fortunately, images were saved documenting that the hematoma must have existed prior to the epidural injection rather than because of the epidural injection. Additionally, by not injecting further, the ILESI did not increase volume to the mass.

Recognition of intravascular uptake and contrast spread is needed to avoid inadvertent injection of medications into the vascularity. Furman et al. [38] reported an overall rate of 11.2% for intravascular injection for lumbosacral TFESIs. TFESIs at the S1 level had an intravascular injection rate of 21.3%, compared with 8.1% for all lumbar injections. With ILESIs, intravascular uptake is more common with needle placement in the lateral portion of the spinal canal than midline because the internal posterior vertebral venous plexus within the epidural space is located predominantly dorsolaterally [39]. The incidence of intravascular uptake with ILESI is reported by Sullivan et al. [40] to be 1.9%. Using a blood “flash back” or blood aspiration to predict an intravascular injection was not reliable. It is postulated that there is not enough pressure in the venous system to result in spontaneous blood flow into the needle hub. However, during contrast or medication injection, there is enough positive pressure to keep these smaller vessels distended, resulting in intravascular injection [38, 40, 41]. Therefore contrast injection is essential to minimize subsequent intravascular injection of medications [38]. Intravascular uptake is twice as likely to occur in patients over 50 years of age [40].

In the studies that reported the incidence of intravascular injections, no complications or adverse affects were found from intravascular, presumably intravenous, injection of contrast, steroid, or local anesthetic. [38, 40, 41]. Theoretically, patients may experience temporary adverse reactions from systemic uptake of local anesthetics. These symptoms include a range of minor symptoms: dizziness, tinnitus, disorientation, muscle twitching, and metallic taste to major symptoms: seizures, unconsciousness, and coma. The severity of symptoms depends on the amount of local anesthetic used [42]. Intravenous uptake of medications may also diminish the efficacy of epidural injections [38]. For example, when performing medial branch blocks, Dreyfuss found a 50% false negative rate with intravascular uptake even when the needle was repositioned such that intravascular uptake was eliminated [43].

While intravenous penetration is relatively common with minimal sequelae, intraarterial penetration and subsequent injection of particulate steroids can lead to a catastrophic outcome. This occurs via the artery of Adamkiewicz which travels with the nerve root through the neural foramen and supplies the anterior spinal artery. This can lead to a spinal infarction and paraplegia [12, 13, 44]. Use of contrast before injection helps to avoid this complication by being able to distinguish an epidural spread versus an intravascular spread. Since the artery enters the spinal canal 85% of the time between T9 and L2 [12, 13] and is located on the left side 63% of the time [14], heightened precaution should be taken for TFESIs at these levels. Houten et al. [12] reported on three cases of paraplegia after lumbosacral nerve block believed to be the result of inadvertent intraarterial injection and an unusually low origin of the artery of Adamkiewicz. In each instance, penetration of the vessel was undetected, i.e., no blood flashes back. In two procedures the needles were placed transforaminally, one at L3–4 on the left and one at L3–4 on the right, and in the third the tip of the needle was placed immediately lateral to the S1 nerve root [12].

Some injectionists utilize digital subtraction angiography (DSA) to help distinguish intraarterial injection of contrast and thus avoid further injection of local anesthetic or steroids into the arterial system. DSA is a computer-assisted X-ray technique that separates and removes images of bone and soft tissue to permit visualization of vascular structures. DSA, however, is not routinely available in most facilities. In our practice, we utilize dexamethasone sodium phosphate for all TFESIs. There has yet to be any reported complications with this preparation. This may be due to its non-particulate nature [45]. Derby et al. [45] found that dexamethasone sodium phosphate particle size is approximately 10 times smaller than red blood cells, the particles do not appear to aggregate, and they have the lowest density compared to other commonly used steroid preparations (e.g., triamcinolone acetonide, methylprednisolone acetate, betamethasone sodium phosphate, and betamethasone acetate). Theoretically, these attributes should lower the risk of embolic infarcts or prevent them from occurring after intraarterial injection [45]. There may be a trade off in the efficacy of this non-particulate steroid in that it has a shorter duration of action. Dreyfuss et al. [46] reported that the effectiveness of dexamethasone was slightly less than that of triamcinolone, but the difference was neither statistically nor clinically significant. Since particulate steroids have been associated with spinal cord infarctions [12, 44, 47], perhaps a safer yet less efficacious non-particulate steroid should be used [46].

In the primary author’s experience, there is a theoretical advantage to the retroneural and retrodiscal techniques in TFESIs in terms of vascular anatomy. Since the optimal target area places the tip of the needle more dorsal in the intervertebral foramen, there is less risk of pinning a spinal artery against the posterior wall of the vertebral body or injecting into it.

Direct trauma to a spinal nerve or dorsal root ganglion by a needle is another complication of inadvertent needle placement, especially when performing TFESIs. Severe pain is caused with this trauma so it is important not to over sedate the patient so that this complication is not masked. If severe pain accompanies needle placement, then the needle should be slightly withdrawn and its position reassessed [18]. In the first author’s experience, injecting a small volume of local anesthetic while repositioning the needle slightly often quickly alleviates the discomfort. Caution should be used such that the anesthetic is not forcibly injected. Furthermore, staying more posterior in the foramen, i.e., the retroneural approach may help avoid nerve trauma. The retrodiscal approach may also reduce the incidence of nerve trauma relative to the classic subpedicular approach in that the nerve may rest further from the target point (Fig. 5). The drawback to the retrodiscal approach may be a less recognizable epidural spread of contrast compared to the other classic approaches. These phenomena, to our knowledge, have not been formally reported or studied.

Inadvertent spinal dural puncture during ILESIs and TFESIs with subsequent entry into the subdural and subarachnoid space can also occur. Dural puncture can occur with ILESIs when the needle is advanced beyond the dorsal epidural space, thereby entering the central spinal canal. Dural puncture has also been reported with TFESIs via penetration of the dural sleeve that surrounds the exiting spinal nerves [48]. Cerebral spinal fluid (CSF) flashback is typically used to recognize the complication of a dural puncture. Recognition of epidural contrast spread versus subdural and subarachnoid contrast spread patterns is essential because dural penetration may not be accompanied by CSF flashback. This is especially true when performing TFESIs [48]. Figure 6 illustrates dural puncture during a TFESI with subdural spread of contrast in a previously reported case study [48].

If local anesthetics are injected intrathecally, blockade of neural elements can result in central canal, cauda equina, and conus medularis syndromes depending on the level of penetration and blockade. Other reported complications with intrathecal injections of medications include persistent parathesias, arachnoiditis, and meningitis. Temporary respiratory depression, ascending weakness/sensory loss, apnea, and unconsciousness may also occur and are felt to be related to ascending subdural spread of anesthetics [26, 48‐52]. The amount of local anesthetic typically used in lumbar epidurals (6–8 ml) usually is not sufficient enough to cause respiratory depression. However, a larger volume within the subdural space can ascend rapidly in a cephalad direction causing serious cardiovascular and respiratory effects [48, 53, 54]. Chauhan et al. [53] describe a case of unintentional combined epidural and subdural block while attempting to perform an epidural block for transurethral resection of the prostate. A 20 ml injection of 1.5% lidocaine and 0.5% bupivacaine resulted in aphonia and respiratory paralysis requiring endotracheal intubation and controlled ventilation for 3 h. The potential for such serious complications heightens the operator’s need to recognize both subarachnoid and subdural contrast spread patterns.

Dural punctures may also lead to spinal headache. These headaches are typically severe, dull, non-throbbing pain, and fronto-occipital in location, that aggravate in the upright position and diminish in the supine position. CSF can leak through the dural puncture leading to a loss of CSF pressure and a drop in brain volume. The mechanism producing the headache is unclear, though two theories do exist. First, in an upright position there is tension on the meninges and other intracranial structures which have nociceptors causing pain [55]. Secondly, it is thought that in the upright position more CSF is forced to exit the dural puncture and the body compensates by venodilation causing pain related to vascular distension [55].

When using air with the “loss of resistance” technique in ILESI to identify the epidural space, there are risks with injecting too much air. One possibility is placing the needle both epidurally and intravascularly without a blood flash back. A subsequent injection of air can then cause an air embolism to develop within the vasculature. MacLean and Bachman [56] depicted a case of syncope, arrhythmia, cardiac ischemia, and neurologic deficit after a spinal epidural injection which caused an arterial gas embolus. There is also the possibility of injecting excessive air epidurally to mimic a mass lesion. Ammirati and Perino [57] documented a case of new neurological symptoms occurring immediately after a lumbar epidural. An MRI revealed trapped air displacing the dural sac. Attention should be paid to the amount of air being injected during epidurals to limit this problem.

ILESI and TFESI can also spread medications to structures outside of the intended epidural target, especially with aberrant or pathological anatomy. Finn and Case [58], for example, recounted a case of disc entry as a complication of transforaminal injection. Though this occurrence is rare, we do not consider this a complication because anecdotally we have found excellent outcomes with injecting around the spinal nerve and into the disc herniation simultaneously. Figure 7 illustrates intradiscal spread of contrast and subsequent medications with a TFESI.

Urinary complications occur more commonly in elderly males, multiparous females, and patients who had inguinal and perineal surgery [22]. The administration of local anesthetics around the lumbar and sacral nerve roots has a higher incidence of urinary retention [59]. Epidural block of S2–S4 root fibers decreases urinary bladder tone and inhibits the voiding reflex [60].

The risk of fluoroscopic exposure to the patient is minimal for one or several isolated fluoroscopic guided epidurals [22]. A properly calibrated digital fluoroscopy machine delivers a low dose of ionizing radiation. It is the cumulative exposure to the physician, nurses, radiology technicians, and anyone else that is routinely involved in these procedures that are at risk for complications [61‐63]. Possible complications are cancer, sterility, cataract development, bone marrow suppression, and skin desquamation. The annual maximum target area/organ permissible radiation doses are as follows: thyroid 50 rem (roentgen equivalent in man), extremities 50 rem, lens of the eye 15 rem, gonads 50 rem, whole body 5 rem [64]. Manchikanti et al. documented the radiation exposure to clinicians after fluoroscopic interventional spinal procedures in 1,000 consecutive patients. The radiation exposure to the physician on the outside of the lead apron over the upper chest was 1,345 mrem for the entire period. Exposure for 2,000 procedures extrapolated to 2,690 mrem outside the apron which is within the annual limits of exposure [63].

There are many ways to limit the amount of exposure to the patient, physician, and staff. The National Council on Radiation and Measurements (NCRP) endorse the concept of ALARA (as low as reasonably achievable), which is based on the premise that all radiation exposures that can be prevented should be prevented [64]. Fishman et al. [65] described three major ways to reduce clinician radiation exposure: (1) maximize the distance from the radiation source, (2) minimize the time of radiation, (3) use protective shielding devices.

Radiation dissipates at the inverse square of the distance from the fluoroscopy tube. Therefore, standing six feet or more from the tube reduces excessive exposure [59‐63]. The fluoroscopy anode should be kept under the procedure table thus the patient absorbs the bulk of the directed radiation [63, 65]. Intermittent fluoroscopy, last image hold, and pulsed fluoroscopy are ways to reduce radiation times [66]. In the lead author’s experience, good technique and a clear understanding of the relevant anatomy and contrast patterns also minimizes radiation exposure by reducing the time to set up the images and perform the injections. According to Fishman, Manchikanti, et al. [63, 65], all staff in the procedure room should at least wear a lead apron and thyroid shield to decrease radiation exposure. Radiation dosimetry badges should be worn to monitor exposure [63, 65]. Manchikanti et al. showed that in 1,000 consecutive procedures the radiation exposure inside the thyroid shield and lead apron was 0 mrem.

Medication complications are rare with the drugs used in lumbar spinal injections. The most common local anesthetics used are lidocaine and bupivacaine. Their primary route of action is reversibly blocking the sodium channels in nerve and muscle membranes [67]. Although very uncommon, physicians must be aware of possible allergy to local anesthetic or its preservative, which occurs within 2 h after epidural injection but has been known to happen up to 6 h later [68, 69]. Inadvertent intrathecal injection of local anesthetic can result in spinal anesthesia as previously addressed.

Corticosteroids inherently have side effects: dizziness, headache, facial erythema, transient hypotension and hypertension, gastritis, mood swings, pruritus, insomnia, and menstrual irregularities [70]. Prolonged therapy with corticosteroids may result in suppression of the hypothalamus–pituitary–adrenal (HPA) axis by inhibiting adrenocorticotropic hormone (ACTH) [70]. Rarely can a Cushingoid syndrome develop with epidural steroid injection [70]. Diabetics can have a temporary elevation of blood sugar levels up to 3–7 days. Patients with congestive heart failure should be aware of possible fluid retention due to corticosteroids, although extremely uncommon [70]. No studies show a relationship between epidural steroid injection and osteoporosis or avascular necrosis [70]. Inadvertent intravascular injection of particulate corticosteroids can cause occlusion of small end arteries as discussed above. All in all corticosteroid use in epidural injections is safe when administered carefully.

Contrast media are usually non-ionic radiocontrast agents with a more physiologic osmolality and less free iodine. However, hypersensitive reactions are still possible and would be evident within the first few minutes after injection. Anaphylactic reactions involve IgE-mediated release of vasoactive substances after exposure to an antigen to which there has been previous exposure and sensitization [71, 72]. A true IgE type I hypersensitivity is rare and only in severe cases [72]. Anaphylactoid reactions occur through a non-immune mechanism. They are pseudoallergic reactions that are caused by the direct release of histamine and other mediators from basophils and mast cells [71, 72]. Clinically, these two reactions are indistinguishable from one another. If an allergy is suspected then pretreatment with antihistamines and corticosteroids should be considered [72, 73] or gadolinium should be used as an alternative [74]. These reactions are highly unusual in extravascular procedures [75].